-
Journal of Constructional Steel Research 128 (2017) 335–353
Contents lists available at ScienceDirect
Journal of Constructional Steel Research
Influence of transverse loading onto push-out tests with
deepsteel decking
Sebastian Nellinger a, Christoph Odenbreit a,⁎, Renata Obiala a,
Mark Lawson ba University of Luxembourg, L-1359
Luxembourg-Kirchberg, Luxembourgb University of Surrey, Guildford,
GU2 7XH, UK
⁎ Corresponding author.E-mail address:
[email protected] (C. Odenb
http://dx.doi.org/10.1016/j.jcsr.2016.08.0210143-974X/© 2016 The
Authors. Published by Elsevier Ltd
a b s t r a c t
a r t i c l e i n f o
Article history:Received 9 July 2015Received in revised form 2
August 2016Accepted 23 August 2016Available online 10 September
2016
This paper presents the results of 20 push-out tests on shear
stud connectors, placed centrally in the ribs of58 mm and 80 mm
deep steel decking. The tests were designed to investigate the
realistic load–slip behaviourof the shear connectors and the
influence of transverse loading. The tests considered two different
stud diametersand the effect of concentric and eccentric transverse
loading. In addition, the influence of a second layer of
rein-forcement, the welding procedure and the number of shear
connectors in each rib have been considered. The ob-served
influence of these parameters on the load–slip behaviour is
presented and explained with regard tomaterial properties and
load-bearing models. In addition, the test results are compared
with the current analyt-ical approaches, which are shown to be
non-conservative in some cases, because the presented deck
shapeswerenot well considered in the development and calibration of
EN 1994-1-1.
reit).
. This is an o
© 2016 The Authors. Published by Elsevier Ltd. This is an open
access article under the CC BY
license(http://creativecommons.org/licenses/by/4.0/).
Keywords:Deep steel deckingPush-out testShear studTransverse
loadingConcrete failure modesMechanical model
1. Introduction
The application of composite beams and slabs has many
advantagesin terms of economic construction of multi-storey
buildings due to theincrease of stiffness and load-bearing capacity
of the structure. Mostcommonly, composite action and transfer of
shear forces between thesteel beam and the slab is ensured by use
of headed shear studs thatare welded to the top flange of the
beams. The current rules in EN1994-1-1 for the analysis of the
shear connector resistances are basedon the failure modes of studs
in solid slabs and do not sufficiently con-sider the load-bearing
behaviour of studs in the ribs of slabs with mod-ern deep steel
decking. Also, the push-out testing procedure, describedin Annex B
of EN 1994-1-1, was originally defined for solid slabs. Thissetup,
when applied to slabs with steel decking, leads to lower
resis-tances and deformation capacity of the shear studs in
comparisonwith beam test results. The paper develops an appropriate
push-testmethod and assesses various test parameters, such as deck
shape,shear connector size, reinforcement pattern and concentric
and eccen-tric transverse loading, which have not been studied
previously.
1.1. Load-bearing behaviour of shear connectors
The load-bearing behaviour of shear studs in solid slabs is
shown inFig. 1. The shear connectors initially transfer the shear
force P by a com-pression force A acting on the concrete. The
compression force A pushes
pen access article under
against the weld collar at a shallow angle β. With increasing
load, theconcrete in front of the stud is damaged and the shear
force moves toa higher position into the stud shank. This leads to
plastic bending andshear deformations. Because of the fixed support
conditions of thehead of the stud, a tension force C develops in
the stud shank. The ten-sion force C is in equilibrium with a
compression cone in the surround-ing concrete. The compression
struts in the concrete activate frictionforces D between the slab
and the steel flange. Finally, failure occurs inthe stud shank
above the weld collar because of combined tensionand shear
forces.
When the shear stud is placed in the deck rib of a composite
slab, theload-bearing behaviour differs from the behaviour of studs
in solidslabs, as shown in Fig. 2. The deck rib geometry has a
strong influenceonto the load-bearing behaviour. In general, two
loading stages can becharacterised by the two load peaks P1 and P2.
The first peak load P1 isreached when the concrete in front of the
stud is crushed and two plas-tic hinges have developed in the stud
shank. At higher slips, the supportconditions of the head of the
stud lead to a back-anchorage effect. Thus,the head of the stud
introduces compression forces into the still intactconcrete section
which are in equilibrium with the tension force C inthe stud shank.
This effect allows the development of a second peakload P2.
Finally, failure occurs in form of a concrete pullout cone orstud
rupture.
The development of this failuremechanism requires a sufficient
em-bedment depth of the head of the stud into the continuous part
of theconcrete slab topping. If the embedment depth is relatively
small, thesupport reaction of the head of the stud cannot be
introduced into theconcrete. In these cases, the concrete fails in
a brittle form and a failure
the CC BY license
(http://creativecommons.org/licenses/by/4.0/).
http://crossmark.crossref.org/dialog/?doi=10.1016/j.jcsr.2016.08.021&domain=pdfhttp://creativecommons.org/licenses/by/4.0/http://dx.doi.org/10.1016/j.jcsr.2016.08.021mailto:[email protected]
logohttp://dx.doi.org/10.1016/j.jcsr.2016.08.021http://creativecommons.org/licenses/by/4.0/http://www.sciencedirect.com/science/journal/0143974X
-
Fig. 1. Load-bearing behaviour of shear studs in solid slabs
according to Lungershausen [1].
Fig. 2. Load-bearing behaviour of shear studs placed in the ribs
of composite slabsaccording to Lungershausen [1].
Fig. 4. Dimensions of the push-out test specimen according to EN
1994-1-1 Annex B [2]and force distribution according to Roik et al.
[3].
336 S. Nellinger et al. / Journal of Constructional Steel
Research 128 (2017) 335–353
mechanismwith only one plastic hinge develops (see Fig. 3).
Therefore,the behaviour of the shear stud is also influenced by the
geometry of thesteel decking and the shear stud itself.
In addition to the stresses that are introduced into the
concrete di-rectly by the shear stud, additional stresses occur
because of the loadingof the concrete slab itself. The loading of
the slab leads to stressesresulting from vertical loads and bending
moments acting on the slabat the line of the shear connectors.
These stresses affect the crushingof the concrete in front of the
stud, as higher stresses can be reachedin multi-dimensional
compression. The embedment conditions of thehead of the stud may
also be influenced by the development of largecracks. These effects
are not yet well investigated and so far not consid-ered in the
push-out test as proposed in EN 1994-1-1 Annex B [2].
Fig. 5. Single push-out test used by Döinghaus [5].
1.2. Test setups to investigate the load–slip behaviour
The push-out test specimen for solid slabs, as given in EN
1994-1-1Annex B2 [2], is shown in Fig. 4. The distribution of the
shear forces ac-cording to Roik et al. [3] is suitable to reflect
the behaviour in real beamswith solid concrete slabs.
However, when deep steel decking is used in concrete slabs, the
ob-tained load–slip behaviour from push-out tests can result in up
to 30%lower stud resistances and lower displacement capacities than
in com-posite beam tests using similar configurations [4].
The load–displacement behaviour of a push-out test is strongly
de-pendent on the boundary conditions of the concrete slab.
Specimens
Fig. 3. Failure of ribs because of a too small embedment depth
of the stud according toLungershausen [1].
with sliding bearings may underestimate the real shear
resistance,whereas for tests with tension ties or rigid horizontal
restraints, theshear resistance may be overestimated [5–7].
The differences of the behaviour of the shear connection in
push-outtests and in beam tests led to the development of
alternative test setupsover recent years, such as the single
push-out test [5] (see Fig. 5) and thehorizontal push-off test
[6,8] (see Fig. 6).
The horizontal push-off test represents a small step towards the
con-sideration of transverse loads because the self-weight of the
slab istaken into account. Other research [4,10] explicitly applied
transverseloads to normal push-out specimens (see Fig. 7).
Typically, concentricloading positionswere used. Currently, the
degree of transverse loadingthat should be applied in these tests
is under discussion. According toHicks and Smith [4], who
investigated transverse loads of 4% to 16% ofthe shear load, a
value of 12% transverse load was suitable to represent
Fig. 6. Horizontal push-off test used by Lam et al. [9].
-
Fig. 7. Transverse loaded push-out test used by Hicks and Smith
[4].
337S. Nellinger et al. / Journal of Constructional Steel
Research 128 (2017) 335–353
the behaviour of the shear studs in the accompanying composite
beamtests.
Fig. 8. Internal forces andmoments at the line of the shear
studs in a composite beam andtheir reflection in a push-out
test.
2. Consideration of transverse loading in the presented
push-outtests
The experimental results of Hicks et al. [4,10] showed an
improve-ment of the shear stud resistance for slabswith steel
decking,when con-centric transverse loads were applied. The
beneficial influence oftransverse loads is not considered in the
assessment of the shear resis-tance of studs according to EN
1994-1-1 [2] or the other design equa-tions presented in this paper
[1,11], and the mechanisms of how thetransverse load influences the
load-bearing behaviour of the shear con-nection, shown in Figs. 1
to 3, have not been established. In addition, theinfluence of the
negative moment of the slabwas not considered. To re-spect these
effects in the presented push-out tests and in a later engi-neering
model of the shear resistance of headed studs [12], transverseloads
were applied with an eccentricity, as shown in Fig. 8c.
To investigate the influence of transverse loading on the
load-bear-ing behaviour of headed studs, it is necessary to define
the degree oftransverse loading for the conduction and evaluation
of the tests. Fig.8a shows the internal forces and moments acting
on a compositebeam and its concrete slab. The structural analysis
of the slab resultsin the vertical forces vR and vL and the
negative bending moment m.The vertical forces vR and vL lead to
compression in the shear interfacebetween the slab and the steel
profile. The transverse load, TL, is thenthe sum of the vertical
forces of the slab according to Eq. (1). The struc-tural analysis
of the composite beam results in internal forces and mo-ments
acting on the slab and the steel section as shown in Fig. 8a.
InSection I acts, the concrete compression force NcI, and in
Section IIacts, the concrete compression force NcII. The difference
of these com-pression forces is the shear force, P, that is
transferred between SectionI and Section II, as shown in Eq. (2).
The degree of transverse loading, ρ,is then the ratio of the
transverse load, TL, to the shear force, P, as shown
in Eq. (3):
TL ¼ vLj j þ vRj j ð1Þ
P ¼ NcII−NcI ð2Þ
ρ ¼ TLP
ð3Þ
where
TL transverse load acting on the shear interface
-
338 S. Nellinger et al. / Journal of Constructional Steel
Research 128 (2017) 335–353
vL, vR vertical forces of the slabNcI concrete compression force
in Section INcII concrete compression force in Section IIP shear
force between steel beam and concrete slab between
Section I and Section IIρ degree of transverse loading
Apreliminary studywas conducted to define reasonable values for
thedegree of transverse loading. A single-span composite beam as
inner sup-port of a two span composite slabwith an imposed load of
qk=3.5 kN/m2
was considered (see Fig. 9). Nominal material properties of
steel grade S355 and concrete gradeC30/37were assumed. Twodifferent
deck shapeswith a deck height of 58 and 80 mmwere considered. The
shear stud re-sistance was calculated according to EN 1994-1-1
[2].
The results of this study are shown in Fig. 10. The degree of
transverseloading, ρ, is mostly dependent on the type and span of
the decking andthe type of shear connector. Deeper decking led to
higher degrees of trans-verse loading because the shear force Pper
studwas smaller than for shallowdecking. Longer beam spans led to
lower degrees of transverse loading be-cause the bending moment,
and hence the required shear force, increased.
For typical spans of slabs with 58mmdeep decking, degrees of
trans-verse loading between 6.6% and 12.2% were obtained, while for
80 mmdeep decking, the values were between 13.0% and 20.8%. Thus,
the trans-verse load of 12% proposed by Hicks and Smith [4] could
not be achievedin all cases considering the loading conditions of
the composite beam.
The transverse loads in the presented test programme were
chosento reflect degrees of transverse loading of 8% and 16%,which
reflect typ-ical values for the two considered deck shapes. To
investigate the effectof the eccentricity, e, the push-out tests
have been conducted with con-centric transverse load application
(e=0, see Fig. 8b) and eccentrictransverse load application (in
which e=380 mm, see Fig. 8c).
3. Test programme and setup
3.1. Test setup for transverse loading
A possible test setup for the application of transverse loads
was de-veloped by Hicks [4], as shown in Fig. 7. This setup has the
benefit that
Fig. 9. Static system used for parametric studies on the degree
of transverse loading.
for each shear interface there was a horizontal hydraulic jack
to applythe transverse load. For the tests presented in this paper,
only one hy-draulic jack was used for the application of the
transverse load. There-fore, a clamping device as shown in Figs. 11
and 12 was used.
The vertical shear load is applied directly to the specimen by
the pri-mary hydraulic jack, as shown in Fig. 11a. Using the
clampingdevice, thehorizontal secondary jack applies the transverse
load to both slabs of thetest specimen. The secondary jack pushes
with the force TL on the ele-ments ‘HP2’ and ‘HP3’ to apply a
concentric transverse load to slab‘S1’, see Fig. 11b. At the same
time, the secondary jack pushes with theforce TL on element ‘HP1’.
Drawbars are used to transfer the force TLfrom element ‘HP1’ to
‘HP5’, see Fig. 11b. This means that the secondaryjack pulls the
elements ‘HP5’ and ‘HP4’ to apply the concentric trans-verse force
TL to slab ‘S2’.
To ensure that only the influence of the transverse loading is
inves-tigated and for eccentric loading the bending of the slabs is
not restraint,the friction at the supports of the specimen is
minimised with pads ofpolytetrafluoroethylene (PTFE).
Some testswere conductedwithout transverse loading for
compara-bility to other data sources. In these tests, the slabs
were placed on amortar bed. A tension tie, as shown in Fig. 1, was
not applied.
The vertical hydraulic jack, which applied the shear force to
thespecimen, was displacement controlled when loading the specimen
tofailure. The horizontal hydraulic jack, which applied the
transverseload, was force controlled. Two different methods were
used to adjustthe applied transverse load during the test:
1. ‘Variable transverse load’For the first test of a series, the
load-bearing capacity of the specimenis unknown. To ensure that the
applied transverse load confirms tothe desired degree of transverse
loading, ρ, the procedure of a vari-able transverse load is used.
In this case, the vertical test load, 2P, iscontinuously measured.
The transverse load, TL, applied by the hori-zontal hydraulic jack
is controlled by a programme that calculatescontinuously the
required transverse load from the currently mea-sured vertical test
load: TL=ρ ⋅P. During the test, the control pro-gramme continuously
adjusts the force applied by the horizontalhydraulic jack according
to the calculated value of the transverseload.
2. ‘Constant transverse load’When the load-bearing capacity of
the specimen can be estimated,for example, from the results of a
test with variable transverse load,the procedure of a constant
transverse load can be used. Based onthe estimated load-bearing
capacity and the desired degree of trans-verse loading, the force
that must be applied by the horizontal jack iscalculated. The
transverse load is applied by the horizontal hydraulic
Fig. 10. Degree of transverse loading, ρ, for different deck
shapes.
-
Fig. 11. Schematic view of the clamping device for the
application of transverse loads.
339S. Nellinger et al. / Journal of Constructional Steel
Research 128 (2017) 335–353
jack before application of the vertical load. During the test,
the hori-zontal jack is set to maintain a constant force:
TL=constant.
3.2. Test programme and material properties
An overview of the test programme is shown in Table 1 and Figs.
13and 14 show the dimensions of the test specimens.
The tests in series 1-04 to 1-06 investigated the use of 58 mm
deepdecking, in which the concrete strength was about 30 N/mm2.
Twolayers of reinforcementwere placed in the slabs, and the
optional recessaccording to EN 1994-1-1 Annex B [2] at the bottom
of the slabs, asshown in Fig. 13, was used. A reinforcement bar of
20 mm diameterwas placed above the recess to prevent vertical
splitting of the speci-men. The steel decking was pre-punched and
single shear studs perrib with a diameter of 22 mm were welded
directly to the flange ofthe beam. The stud height after welding
was measured at about124 mm, which results in an embedment depth of
about 3.5 diametersinto the concrete above the decking. Thus, the
required minimum em-bedment of 2 diameters, according to EN
1994-1-1 [2], was satisfied.
Series 1-04 and 1-05 investigated the influence of concentric
trans-verse loads. According to the results of the study presented
in Section2, the slabs had a transverse load of about 4% of the
test load in series1-04 and about 8% of the test load in series
1-05. Thus, the degree oftransverse loading at each shear interface
was about 8% and 16%.
Series 1-06 investigated the influence of the negative bending
mo-ment in the slab at a transverse load of about 4% of the test
load. The de-gree of transverse loading at the shear interface
reflects the typical valueof about 8%, which was found in the study
presented in Section 2. Thetransverse load was applied with an
eccentricity of 380 mm.
The tests in series 1-09 toNR1used 80mmdeep decking. These
testsused a single layer of reinforcement, except for series 3-01,
which hadtwo layers of reinforcement. Pairs of shear connectors at
a transversespacing of 100 mmwere welded through the decking in
series 1-09 to3-01. Series NR1 had single studs per rib that were
welded throughthe decking and a single layer of reinforcement was
placed in the slab.The shear connectors were 19 mm diameter, and
the height afterwelding was about 119 mm for through deck welded
studs and about121 mm for studs welded directly to the beam. Thus,
the embedmentdepth of the head of the stud into the concrete above
the decking was
-
Fig. 12. Test setup for the application of transverse load
assembled for concentric loading.
340 S. Nellinger et al. / Journal of Constructional Steel
Research 128 (2017) 335–353
2 to 2.2 diameters. All specimens with 80 mm deep decking were
con-ducted without the recess at the bottom of the slab, as shown
in Fig. 14.
The tests in series 1-09, 1-10 and NR1 investigated the
influence ofconcentric transverse loading, while the influence of
the eccentricitywas investigated in series 1-11.
For all specimens, the concrete was cast in a horizontal
position toreflect the real conditions. The two halves of the
specimenwereweldedtogether prior to testing. The cylinder strength
and Young's modulus ofthe concreteweremeasured at 28 days and at
the day of the test respec-tively and are shown in Table 1. The
tensile strength, fu, of the shear con-nectors was measured as 551
N/mm2.
4. Observed load–slip curves and failure modes
4.1. General results of tests with 58 mm deep decking
Typical load–slip curves for the tests with 58mmdeep
deckingwithconcentric and eccentric transverse loads are shown in
Fig. 15. All testsexhibited a load-bearing mechanism with two
plastic hinges as de-scribed by Lungershausen [1]. All tests showed
a second load peak, butthis peak decreasedwhen eccentric transverse
loadwas applied. Identi-cal specimens without transverse loading,
presented in [13], showed asimilar behaviour with two load peaks.
The observed failure modeswere:
• Rib punch-through, see Fig. 16• Concrete pullout, see Fig. 18•
Stud failure, see Figs. 17 and 21a
4.2. General results of tests with 80 mm deep decking
Examples of typical load–slip curves for
specimenswith80mmdeepdecking are shown in Fig. 19. The observed
failure modes were
• Rib pry-out, see Fig. 20• Stud failure, see Fig. 21b
Fig. 19 shows that there was no significant difference between
testswith single studs per rib (NR1-1) and pairs of shear
connectors (1-10-3)within the first 20 mm of slip. The load–slip
curves were linear until abrittle failure of the concrete ribs
occurred. This led to an immediateloss of stiffness and a drop-off
in the test load by up to 15%. The static re-sistance of the tests
Pe , static was determined at failure. Typically, thestuds
developed only a plastic hinge at the bottom of the stud shank(see
Fig. 20c), except for tests with high transverse compression.
Thebehaviour after the failure of the ribs was strongly dependent
on thetransverse loading, as shown by the comparison of specimens
1-10-1and 1-10-3, shown in Fig. 19. In addition, the boundary
conditions of
-
Fig. 13. Typical dimensions of specimen with 58 mm deep decking.
Fig. 14. Typical dimensions of specimen with 80 mm deep
decking.
Table 1Specimen geometry, concrete strength and loading
conditions of push-out tests.
Series No.
Decking Studs Concrete properties Reinforcement Transverse
Load
hp bm t d hsc nr fc Ec
Bottom Top
∑ V e
[mm] [mm] [mm] [mm] [mm] [−] [N/mm2] [N/mm2] [kN/slab] [mm]
1-04 1 58 81.5 0.88 22.2 124.3 1 30.6 20,900 Q188A Q335A 4.1%+
02 22.2 124.0 30.9 21,500 12.5 03 22.2 124.0 30.9 21,500 12.5 0
1-05 1 58 81.5 0.88 22.2 124.0 1 30.7 22,100 Q188A Q335A 8.2%+
02 22.2 123.8 30.7 22,100 25.0 03 22.2 123.9 32.6 22,800 25.0 0
1-06 1 58 81.5 0.88 22.2 123.6 1 29.9 21,200 Q188A Q335A 4.1%+
3802 22.2 124.1 31.1 21,400 12.5 380
1-09 1 80 137.5 0.90 19.1 118.8 2* 42.6 28,000 Q188A – 8.8 01-10
1 80 137.5 0.90 19.1 118.6 2* 42.6 28,000 Q188A – 17.5 01-10 2 80
137.5 0.90 19.1 118.1 2* 42.6 28,000 Q188A – 13.2 01-10 3 80 137.5
0.90 19.1 118.2 2* 42.6 28,000 Q188A – – –1-11 1 80 137.5 0.90 19.1
119.4 2* 42.6 28,000 Q188A – 3.8%+ 380
2 19.1 118.7 2* 42.6 28,000 17.5 3803 19.1 118.6 2* 42.6 28,000
17.5 380
3-01 3 80 137.5 0.90 19.1 118.3 2* 40.4 26,800 Q188A Q335A –
–3-02 1 80 137.5 0.90 19.1 123.4 2 42.6 28,000 Q188A – – –NR1 1 80
137.5 0.90 19.1 121.3 1* 44.1 25,600 Q188A – – –NR1 2 80 137.5 0.90
19.1 121.2 1* 45.7 25,600 Q188A – 8.8 0NR1 3 80 137.5 0.90 19.1
121.0 1* 44.7 25,600 Q188A – 17.5 0
hp: height of the deck profile.bm: width of the decking at
0.5hp.t: deck thickness.d: measured stud diameter.hsc: measured
stud heightnr: number of studs per deck ribfc: concrete cylinder
strengthEc: measured Young's modulus of concrete∑ V: applied
transverse load.e: eccentricity of transverse load*Welded through
the decking+Percentage of the total test load, permanently
adjusted
341S. Nellinger et al. / Journal of Constructional Steel
Research 128 (2017) 335–353
-
Fig. 15. Typical load–slip curves for push-out tests with 58 mm
deep decking withconcentric transverse loading (TL) and eccentric
transverse loading (TL).
Fig. 17. Stud deformation after concrete pullout.
342 S. Nellinger et al. / Journal of Constructional Steel
Research 128 (2017) 335–353
the test had a significant influence on the load–slip behaviour
as report-ed in [7]. For specimens NR1-1 and NR1-3, failure of the
shear studs wasobserved at about 17mmslip. The studs failed in
theweld collar as therewas high porosity in theweld (see Fig. 21b).
For specimenswith pairs ofshear connectors, stud failure was not
observed.
Load–slip curves similar to the tests with 80 mm deep steel
deckingwere obtained by Hicks and Smith [4], where shear connectors
with anominal height of 100mmwerewelded in the ribs of 61mmdeep
deck-ing. In these tests, the studs did not satisfy the minimum
embedmentdepth of 2 diameters, which is required by EN 1994-1-1
[2]. Thus, thesupport reaction of the head of the stud, which would
be necessary forthe development of the upper plastic hinge, could
not be introducedinto the concrete slab and a failuremodewith only
one plastic hinge de-veloped (see Fig. 20c).
Hicks and Smith [4] referred to the failure of the ribs as
concrete pull-out. Because of the low slip at failure, the force
that would have beennecessary to fail the rib of the slab in
tension could not have developedin the stud shank. In tests with 58
mm deep decking, concrete pulloutwas observed at very large slips
and appeared as a slowly progressingand ductile failurewith two
plastic hinges (see Figs. 15, 17 and 18). Fail-ure of the ribs in
testswith 80mmdeep decking occurred at a very smallslip. Hence, the
loading situation at failure was different, and bending
Fig. 16. Rib punch-through failure.
and shear must have been the dominant loads – instead of tension
–leading to different stud deformations and load–slip curves.
Therefore,this failure is not treated as concrete pullout but
referred to as rib pry-out, as shown in Fig. 20.
5. Evaluation of the test results according to EN 1994-1-1 Annex
B2
EN 1994-1-1 Annex B2 [2] gives a simplified procedure to
determinethe characteristic resistance PRk and slip capacity δuk
out of the results ofpush-out test. The pre-condition for the
application of this procedure isto have a series of three tests
with identical nominal properties. Theshear stud resistance of each
test must not deviate by N10% from themean value of the shear stud
resistance for the series.
Because the tests with 80 mm deep decking varied the degree
oftransverse loading between 0% and about 10% within some
series,they may not be assumed to have identical nominal
properties. For in-formation, Table 2 shows the results of the
evaluation according to EN1994-1-1 Annex B2, even if there are b3
tests with identical nominalmaterial properties or loading
conditions.
Fig. 18. Concrete pullout failure for the bottom rib and rib
punch-through and stud failurefor the top rib.
Image of Fig. 17Image of Fig. 18
-
Table 2Characteristic resistance PRk and slip capacity δukof
push-out tests according to EN1994-1-1 Annex B2.
Series i
Pe Pe,static PRk δu δuk
[kN] [kN] [kN] [mm] [mm]
1-04 1 295.8 281.8 234.1 67.6 55.32 294.4 278.5 61.53 274.6
260.1 68.7
1-05 1 279.8 265.6 239.0 62.1 51.62 316.6 278.5 61.63 297.7
283.4 57.4
1-06 1 292.4 275.6 236.5 71.0 59.42 280.5 262.8 66.0
1-09 1 338.2 312.9 281.6 7.6 6.81-10 1 365.1 326.0 293.4 25.7
23.21-10 2 371.5 314.3 282.9 5.1 4.61-10 3 283.7 249.7 224.8 5.0
4.51-11 1 437.1 411.3 29.1 16.9 15.2
2 421.4 368.3 20.53 394.6 365.7 22.3
3-01 3 422.2 378.5 340.7 2.4 2.23-02 1 296.0 279.2 251.3 1.8
1.6NR1 1 316.2 287.4 258.6 3.7 3.3NR1 2 300.0 271.7 244.5 5.9
5.3NR1 3 281.8 259.5 233.5 3.3 2.9
Pe: experimental resistancePe,static: static resistance.PRk:
characteristic resistance for the seriesδu: displacement
capacityδuk: characteristic displacement capacity
Fig. 19. Typical load–slip curves for push-out tests with 80mm
deep decking, single studs(n = 1) and pairs of studs (n = 2)
without transverse loading (no TL) and concentrictransverse loading
(TL).
Table 3Comparison of test results with 58 mm deep decking and
different degrees of concentrictransverse loading.
1-03[13] 1-04 1-05
∑ V [kN/slab] 0 12.5 25.0fc [N/mm2] 42.5 30.8a) 32.3f c
f c;ref[−] 1.38 1.00 1.02
Pe [kN] 363.4 288.3b) 298.0Pe
Pe;ref[−] 1.26 1.00 1.03
δu [mm] 44.3 65.9b) 60.4δu
δu;ref[−] 0.67 1.00 0.92
a)Reference concrete strength fc,refb)Reference resistance
Pe,refc)Reference displacement capacity δu,ref
343S. Nellinger et al. / Journal of Constructional Steel
Research 128 (2017) 335–353
The characteristic resistance PRk of a test series is
theminimumvalueof all three tests reduced by 10%. The
characteristic resistance PRk,shown in Table 2, was derived from
the static resistances Pe ,static.
For tests with 58 mm deep decking, the characteristic resistance
ofthe specimen was between 234 and 239 kN, which led to a
resistanceper shear stud of 58.5 to 59.8 kN.
Tests with 80mmdeep decking and pairs of studs showed a
charac-teristic resistances of 225 to 341 kN—i.e. 28.1 to 42.6
kN/stud. The largescatter was related to the brittle type of
concrete failure. The resistanceimproved for higher transverse
loads and a second layer of reinforce-ment. For single studs, the
characteristic resistance was between 233and 258 kN -i.e. 58.4 to
64.7 kN.
According to EN1994-1-1Annex B2 [2], the displacement capacity
δuof ashear stud is the slip atwhich the test loaddrops for
thefirst time to the char-acteristic resistance PRk. The
characteristic slip capacity δuk is the lowest valueof δu obtained
for the test series reduced by 10%, as shown in Fig. 22. In
thepresented evaluation, the influence of relaxation is not
considered for the de-termination of the displacement capacity. The
results are shown in Table 2.
All specimens with 58 mm deep decking showed a very ductile
be-haviour with characteristic slip-capacities δuk between 51 and
59 mm.The slip capacity appeared to be largely influenced by the
degree andposition of the transverse load.
For 80 mm deep decking, the characteristic slip-capacities
weresmaller. In most cases they did not satisfy the 6 mm criterion
of EN1994-1-1 [2] to be classified as ductile. However, the general
shape ofthe load–slip curves (see Fig. 19) should allow the
assumption of a duc-tile behaviour in most cases. The problem with
the defined slip-capaci-ties arose as the brittle concrete failure
led to a relatively large butlocalised drop-off in the test load
that is related to a change of theload-bearing behaviour. The
application of further displacement typi-cally led to a slow and
ductile decrease of the load.
6. Discussion of influencing parameters
6.1. Influence of variable versus constant transverse
loading
For the first tests in series 1-04 to 1-06 and series 1-11, the
appliedtransverse load was continuously measured and re-adjusted
to
maintain the percentage of the test load given in Table 1. In
all othertests, the transverse load was applied with constant
values as shownin Table 1. For example, the load–slip curves of
series 1-04 and 1-11are shown in Fig. 23.
Comparing the load–slip curves of tests with variable
transverseloads with those of tests with constant transverse loads,
no significantinfluence of the loading procedure on the initial
stiffness of the shearconnectors and the failure load was
observed.
6.2. Considerations on the multi-axial stress state for 58 mm
deep decking
Examples of load–slip curves for 58mmdeep deckingwith
differentdegrees of concentric transverse loading are shown in Fig.
24. The re-sults of series 1-03 [13] and 1-05 are compared with
series 1-04 inTable 3. The comparison of series 1-04 and 1-05 shows
that highertransverse loads led to an increase of the test load Pe
of 3% and a de-crease of the displacement capacity δu of 8%. The
tests reported byEggert at al. [13], which had an approximately 12
N/mm 2 higher con-crete strength, showed an increase of the test
load of 26% and a decreaseof the displacement capacity of 33%.
The similarity of the effect of transverse load and concrete
strengthcan be explained with the multi-dimensional compression
stress state,which increases the failure stress of the concrete
[14] (see Fig. 25).
To investigate the influence of multi-dimensional stress states,
theconcrete strength according to the failure curve of Kupfer et
al. [14]was considered in the determination of the analytical
resistance for con-crete failure according to EN 1994-1-1 [2]. The
concrete strength fc wasreplaced with the increased concrete
strength σ1, as shown in Eq. (4).The stress σ2, which was used to
determine σ1, was assumed as thecompression stress at the shear
interface out of the transverse load
Image of Fig. 19
-
Table 4Comparison of the analytical resistance for different
degrees of transverse loading underconsideration of the
two-dimensional stress conditions acc. to Kupfer et al. [14].
Series
fcσ2f c
σ1f c
PRm,c
Eqs. (8), (10), (7)
[N/mm2] [−] [−] [kN]
1-04a) 30.8 0 1 529.71-05a) 31.3 0 1 534.41−051−04
1.02 – – 1.01
1-04b) 30.8 0.013 1.025 538.41-05b) 31.3 0.025 1.050
552.11−051−04
1.02 1.923 1.024 1.03
σ1: Increased concrete strength acc. to Kupfer et al. [14].σ2:
Transverse compression acc. to Eq. (6).a)Analysis without
consideration of multi-axial stresses.b)Analysis with consideration
of multi-axial stresses.
Table 5Comparison of test results with 80 mm deep decking and
different degrees of concentrictransverse loading.
Test
∑ V PePe
Pe;V¼0 δuδu
δu;V¼0
[kN] [kN/slab] [−] [mm] [−]
1-10-3 0.0 283.7 1.00 5.0 1.001-09-1 8.8 338.2 1.19 7.6
1.511-10-2 13.2 371.5 1.31 5.1 1.011-10-1 17.5 365.1 1.29 25.7
5.11NR1-1 0 316.6 1.00 3.7 1.00NR1–2 8.8 300.0 0.95 5.9 1.59NR1-3
17.5 281.9 0.89 3.3 0.89
344 S. Nellinger et al. / Journal of Constructional Steel
Research 128 (2017) 335–353
(see Eq. (6)):
Pc ¼ 0:374 � α � d2ffiffiffiffiffiffiffiffiffiffiffiffiffiffiσ1
� Ec
pð4Þ
PRm;c ¼ ktPc ð5Þ
σ2 ¼ ∑ Vn � bF � bu ð6Þ
where
Pc concrete failure load in solid slabkt reduction factor in EN
1994-1-1PRm ,c concrete failure with steel decking∑ V transverse
load applied to the slabn=2 number of deck ribs per shear
interfacebF width of steel flangebu bottom with of deck ribα=1.0
for hsc/dN4α ¼ 0:2 � ðhscd þ 1Þ for 3≤hsc/d≤4
Table 6Influence of the eccentricity on the resistance Pe and
the displacement capacity δu.
Series Test
hp ∑V e Pe δu
[mm] [kN] [mm] [kN] [mm]
1-04 1 58 4.1% 0 295.8 67.62 58 12.5 0 294.4 61.53 58 12.5 0
274.6 68.7avg. 58 12.5 0 288.3 65.9
1-06 1 58 4.1% 380 292.4 71.02 58 12.5 380 280.5 66.0avg. 58
12.5 380 286.5 68.5
1−061−04
0.99 1.04
1-10 1 80 17.5 0 365.1 25.71-11 1 80 3.8% 380 437.1 16.9
2 80 17.5 380 421.4 20.53 80 17.5 380 394.6 22.3avg. 80 17.5 380
417.7 19.9
1−111−10
1.14 0.77
The concrete's Young's modulus Ec in Eq. (8) was assumed to be
ac-cording to Eq. (7),whichwas also assumed by Roik et al. [15] in
the eval-uation of Eq. (4). For the presented analysis, the
concrete strength fcwasreplaced with σ1 when Eq. (7) is
evaluated:
Ec ¼ 9500 � f 1=3c ð7Þ
The comparison was performed with the averaged dimensions
andmaterial properties of series 1-04 and 1-05. The resistancewas
calculat-ed with and without consideration of the two-dimensional
failurecurve. The obtained analytical resistances PRm ,c are
summarised inTable 4.
Without consideration of the two-dimensional stress conditions,
theanalytical resistance PRm ,c of series 1-05 is about 1% higher
than for se-ries 1-04 because of the scatter in the measured
material propertiesand dimensions. The simplified assumption of a
two-dimensional stresscondition according to Kupfer et al. [14] led
to an about 3% higher ana-lytical resistance for series 1-05
compared to series 1-04. In fact, an in-crease of the averaged
measured resistances Pe of about 3% wasobserved in the tests (see
Table 3). Thus, for 58 mm deep decking, theobserved influence of
concentric transverse loading on the load–slip be-haviour is mostly
related to the change of the stress conditions in theconcrete. The
stress conditions can be considered by
multi-dimensionalmaterial-laws for the concrete. Nevertheless,
comparing the test resultsof Eggert et al. [13] to series 1-04 and
1-05 clearly shows that the effectof a higher concrete strength
governs the load–slip behaviour morethan the increase of the
concrete compressive resistance for multi-axial stress states.
6.3. Influence of the degree of concentric transverse loading
for 80mmdeepdecking
For the tests with 80 mm deep decking, the effect of the
transverseload was more important than for the shallower decking.
Especiallyfor tests with pairs of studs, the transverse load
strongly improved theload–displacement behaviour (see Fig. 26).
This was valid for the failureload of the ribs, as well as for the
post-failure behaviour.
For the failure load, an increase of about 30% was observed due
totransverse loading (see Table 5). The dependency between the
failureload and the applied transverse loads appeared to be linear
until a trans-verse load of about 13 kNwas exceeded. For higher
transverse loads, thefailure load did not increase further (see
Fig. 28).
The increase of the failure load and post-failure behaviour was
notobserved in tests with only one shear stud per rib (see Table 5
and Fig.27). For single studs per rib, the failure load of the ribs
decreased ap-proximately linearly with the transverse load by up to
11% (see Fig. 28and Table 5). Comparing the load–slip curve without
transverse loadto the load–slip curve with a small transverse load
of 8.8 kN, no signifi-cant difference of the behaviourwas observed
(see Fig. 27). For a highertransverse load of 17.5 kN, the test
load after rib pry-out failure wasabout 50 kN lower than without
transverse load until stud failure wasobserved.
The diversity of the post-failure behaviour for pairs of studs
and sin-gle studs per rib can be explainedwith the
load-bearingmodel shown inFig. 29. Failure is initiated along the
surface A-B-C at point A at a slip ofabout 1 to 2.5 mm. For further
loading, the majority of the loadmust beintroduced into the slab
along the surface B–C. Due to the inclination ofthis surface, a
shear forces T and a normal force N act on the face B–C.The force N
pushes the slab upwards and causes bending moments inthe slab—as
observed by the deformation of the slabs in most push-out tests
(see Fig. 36a). The whole rib rotates around the base of the
-
Fig. 20. Concrete failure surface and stud deformation for rib
pry-out failure.
345S. Nellinger et al. / Journal of Constructional Steel
Research 128 (2017) 335–353
stud and the plastic hinge above the weld collar develops. The
head ofthe stud is supported by the compression strut D. At large
slips, plasticbending deformation in the upper stud shank may
develop if the bear-ing capacity of the compression strut D and the
face B–C are sufficient.The application of a concentric transverse
load restrains the displace-ment of the slab and increases the
force N linearly to the transverseload. For pairs of studs, a
higher shear force P was observed with in-creasing transverse
loads. This effect is limited by crushing of theconcrete below the
surface B–C or the bearing capacity of the compres-sion strut
D.
For single studs, the shear force P is not distributed over
severalstuds and consequently the compression strut D is more
highly loaded.Also, the surface B–C is smaller than for pairs of
studs. Therefore, theconcrete below the face B–C crushes at a lower
shear force P. The
Fig. 21. Comparison of failure surfaces for shea
transverse load acting on the slab must be transferred through
the ribinto the flange of the beam. This changes the compression
strut D andthe surface B–C in addition, which leads to a decrease
of the load-bear-ing capacity for the shear force P when single
studs per rib were used.
6.4. Influence of the eccentricity of transverse loads with 58
mm deepdecking
In testswith 58mmdeep decking, no influence of the eccentricity
onthe first peak load Pe could be observed and the displacement
capacityδu slightly increased by about 3.9% (see Table 6).
Comparisons of theload–slip curves for concentric and eccentric
transverse loading showedthat the second peak load decreased for
the eccentric loaded tests byabout 50 kN (see Fig. 30). In
addition, the final failure for concentric
r studs with different welding procedures.
-
Fig. 24. Comparison of load–slip curves for tests with different
degrees of concentrictransverse loading for 58 mm deep decking.
Fig. 22.Determination and δuk obtained from the load–slip curves
according to EN 1994-1-1 Annex B2 [2], shown for series 1-04.
346 S. Nellinger et al. / Journal of Constructional Steel
Research 128 (2017) 335–353
transverse loading was stud failure – in most cases – and
concrete pull-out. The failure mode observed in the eccentrically
loaded tests was al-ways concrete pullout. According to
Lungershausen [1], thedevelopment of the second peak load is
related to the tension force inthe stud shank.
The influencing parameters for the observed reduction of the
secondloadpeakwhen eccentric transverse loadingwas applied are
cracking of
Fig. 23. Comparison of tests with constant transverse load and
transverse loadmaintainedrelative to the test load.
the concrete and the larger tension force in the studs. Larger
tensionforcesmay occur as the slab slightly lifts at the line of
the shear studs be-cause of the bending deformation of the slab.
The head of the shear studrestrains this up-lift and so the stud is
loadedwith an additional tensionforce.
6.5. Influence of the eccentricity of transverse loads with 80
mm deepdecking
Comparing the load–slip curves for concentric and eccentric
loadedspecimen with 80 mm deep decking (see Fig. 31), no
significant influ-ence on the general behaviour could be
identified.
All three eccentric loaded specimens showed changes of their
stiff-ness at about 300 to 350 kN, but cracking of the ribs and the
drop-offin the test load, which typically occurred after rib
pry-out failure, havebeen observed at test loads of 390 kN to 430
kN. These observationsled to the failure loads Pe reported in Table
2. The identified failureloads of eccentric transverse loaded
specimens were higher than forthe concentric transverse loaded
specimen.
Based on the load-bearingmodel shown in Fig. 29, it can be
assumedthat the eccentricity has no significant influence on the
post-failure
Fig. 25. Failure curves for two-dimensional stress conditions in
concrete according toKupfer et al. [14].
Image of Fig. 22
-
Fig. 26. Comparison of load–slip curves for tests with different
degrees of concentrictransverse loading and pairs of studs in 80 mm
deep decking.
Fig. 27. Comparison of load–slip curves for tests with different
degrees of concentrictransverse loading and single studs per rib in
80 mm deep decking.
Fig. 29. Post-failure behaviour for rip pry-out failure.
Fig. 30. Comparison of load–slip curves for tests with
concentric and eccentric transverseloading for Cofraplus 60.
347S. Nellinger et al. / Journal of Constructional Steel
Research 128 (2017) 335–353
behaviour, because the failure cone is not influenced by
cracking of theconcrete topping. It is not possible to tension the
shear stud due tobending of the slab and the failure surface allows
only compressionand shear forces to be transferred at the face
B–C.
6.6. Influence of through deck welded studs with 80 mm deep
decking
Fig. 32 shows shear studs thatwerewelded through the decking
andshear studs that were welded directly to the flange of the beam
whenpre-punched decking was used.
Fig. 28. Test load plotted versus the transverse load for tests
with 80 mm deep deckingwith pairs of studs per rib and single studs
per rib.
Fig. 33 shows the load–slip curve of a push-out test with shear
studswelded through the decking in comparison with a test with a
pre-punched decking.
It can be seen that the welding procedure had no influence on
thefailure load, because rib pry-out failure occurred in both cases
at a testload of about 280 kN to 300 kN. The test with the
pre-punched deckingshowed a reduced performance for the
post-failure behaviour. The loaddid not increase again until the
steel sheeting came into direct contactwith the weld collar. On the
other hand, through deckwelded studs im-mediately activate the
decking as a tension tie. An additional compo-nent for the shear
force can be transferred by this tension effect of the
Fig. 31. Comparison of load–slip curves for tests with
concentric and eccentric transverseloading (TL) for 80 mm deep
decking.
-
Fig. 32. Through deck welded studs and studs welded directly to
the beam with pre-punched decking.
Fig. 33. Comparison of the load–slip curves for pairs of studs
welded through the deckingand welded directly to the flange of the
beam. Fig. 35. Comparison of the load–slip curves for tests with
one and two layers of
reinforcement.
348 S. Nellinger et al. / Journal of Constructional Steel
Research 128 (2017) 335–353
decking. The tension force is introduced into the slab at the
re-entrantstiffeners on top of the sheeting (see Fig. 11). Because
of the tension ef-fect, a larger shear force was observed for
through deck welded studs.
Fig. 34. Position of reinforcement for tests w
For through deck welded shear studs, the test load increased to
asecond load peak of about 300 kN at approximately 25 mm slip.
Thispeak developed because the tension effect. The drop-off in the
load
ith one and two layers of reinforcement.
-
Fig. 36. Deformation of the push-out test specimen after
testing.
349S. Nellinger et al. / Journal of Constructional Steel
Research 128 (2017) 335–353
which followed was because of the debonding of the decking's
re-en-trant stiffeners from the concrete slab.
It can be assumed that shear studs with pre-punched decking
showan in general 10% lower resistance than shear studs welded
through thedecking (see Fig. 33).
6.7. Influence of the number of reinforcement layers
For the specimen with two reinforcement layers, the bottom
layerwas placed 15mmabove the decking, and for the specimenwith one
re-inforcement layer, it was placed 30 mm above the decking as
shown inFig. 34.
The number of reinforcement layers had a large influence on
thestiffness and the failure load of the ribs, but not onto the
post-failure be-haviour (see Fig. 35).
Rib pry-out failure occurred in the test with a single
reinforcementlayer at a load of 287 kN and a slip of 1.4 mm. For
the test with two re-inforcement layers, rib pry-out failure
occurred at a load of 422 kN andonly 0.9 mm slip. This was an
increase of the failure load of 47%. Thestiffness of the shear
connection was more than doubled. After the fail-ure of the ribs,
the load finally dropped to about 270 kN independentlyof the number
of reinforcement layers.
Fig. 37. Comparison of load–slip curves of tests with 80 mm deep
decking with singlestuds and pairs of studs per rib and no
transverse load.
The lower position of the bottom reinforcement in specimen
3-01-3improved the embedment conditions of the shear stud because
the bot-tom reinforcement layer overlappedwith the failure surface
of the con-crete cone. In addition, two layers of reinforcement led
to a higherbending resistance and bending stiffness of the slab.
Both details con-tribute to the increase of the failure load.
The influence of the number of reinforcement layers on the
displace-ment behaviour of the slabs is shown in Fig. 36. For the
specimen withonly one layer of reinforcement, the slabs were
subjected to higherbending displacements (see Fig. 36a).With two
layers of reinforcement,the slabs did not bend but tilted over and
rotated outwards at the sup-port (see Fig. 36b). In both cases, the
slabs were subjected to horizontaldisplacements because the force
N, shown in Fig. 29, pushed the slaboutwards when the failure
surface slides along the concrete topping.As this global
displacement was not restrained, there was no significantdifference
in post-failure behaviour (see Fig. 35).
6.8. Influence of the number of shear studs per rib
The influence of the number of studs per rib was only
investigatedfor the 80mmdeep decking. As shown in the
considerations on concen-tric transverse loading in Section 6.3,
the number of shear studs per rib
Fig. 38. Comparison of themeasured resistances Pewith the
analytical resistances PRm acc.to EN 1994-1-1 [2].
-
Fig. 39. Comparison of the measured resistance Pe with the
analytical resistance PRmaccording to Lungershausen [1].
Fig. 41. Comparison of the ratios Pe/PRm for different analysis
methods.
350 S. Nellinger et al. / Journal of Constructional Steel
Research 128 (2017) 335–353
determined if the concentric loading was beneficial or not for
this typeof decking.
When considering tests without transverse loading, there was
onlynegligible difference of the load–slip behaviour observed
between sin-gle studs and pairs of studs per rib within the first
15 mm of slip (seeFig. 37). For single studs aswell as for pairs of
studs per rib, the observedfailure mechanism was rib pry-out. The
failure surfaces did not differmuch from each other because the
transverse spacing of the shearstuds was small in comparison to the
width of the failure cone (seeFig. 20). Therefore, the failure load
of the ribs did not vary significantly.
Once a slip of about 15 to 20mmwas reached, significant
differenceswere observed. For single studs per rib, steel failure
occurred because ofthe higher loading per shear stud (see Fig. 28).
In addition, there was noload peak due to the tension component of
the decking because theshear studs punched through the decking.
7. Proposed testing procedure for push-out specimenswith
compos-ite slabs
The presented push-out tests show that the influence of
transverseloading is negligible in practice if the embedment depth
of the head ofthe stud into the concrete topping is high.
However, for studs with a relatively small embedment depth,
trans-verse loading significantly improved the load–slip behaviour
and led toup to 30% higher shear resistances (see Table 5). No
relevant influenceof the loading procedure – constant or variable
transverse loading –and the eccentricity was observed in the
tests.
Fig. 40. Comparison of the measured resistance Pe with the
analytical resistance PRmaccording to Konrad [11].
Based on these observations, it is recommended to conduct
push-outtests as follows:
• With concentric transverse loading, when the embedment depth
ofthe stud is too small to ensure double curvature of the headed
studs.
• Without transverse loading, when the embedment depth of the
studis large enough to ensure double curvature of the headed
studs.
As observed in the tests, the 2 diameters criterion of EN
1994-1-1 [2]is not sufficient to differentiate between a small and
a large embedmentdepth. Amore suitable criterion is given by Konrad
[11]. Konrad numer-ically investigated the influence of the ratio
of the stud height to thedeck height hsc/hp onto the reduction
factor kt. It was found that the cor-relation curves for kt changed
at a ratio of hsc/hp=1.56. This can beinterpreted as a change of
the failure mechanism and correspondswell with the observations in
the presented push-out tests.
According to the results of the studypresented in Section 2, a
conser-vative value for the transverse load of 5% of the total test
load is recom-mended. This is a degree of transverse loading per
shear interface of10%, which is slightly less than the value
proposed by Hicks and Smith[4].
In general, the load-bearing capacity of the test specimen is
notknown. In this case, the procedure of a ‘variable transverse
load’ shallbe used. This means that during the test the transverse
load must bepermanently adjusted to maintain a value of 5% of the
current verticaltest load.
For a series of testswith nominal identical properties, variable
trans-verse loading shall be used for the first test of the series.
A value of 5% ofthe load-bearing capacity of the first test may be
applied as a constanttransverse load in further tests.
Table 7Maximum reduction factors kt,max according to EN 1994-1-1
[2].
nr
tThroughwelded Punched holes
[mm] d ≤ 20 mm 19mm≤d≤22mm
1 ≤1.00 0.85 0.75N1.00 1.00 0.75
2 ≤1.00 0.70 0.60N1.00 0.80 0.60
-
351S. Nellinger et al. / Journal of Constructional Steel
Research 128 (2017) 335–353
8. Comparison with analytical resistances
8.1. Comparison with analytical resistances according to EN
1994-1-1 andRoik et al.
EN1994-1-1 [2] assumes as basic failuremodes either a failure of
thestud (see Eq. (9) [15]) or a concrete compression failure
directly in frontof the weld collar (see Eq. (8) [15]). Both are
failure modes of shearstuds in solid slabs (see Fig. 1). To obtain
the failure load of studs inslabswith trapezoidal decking, this
resistance ismultipliedwith the fac-tor kt according to Eq. (10),
which is assumed to be the mean value.Thereby, the reduction factor
kt is limited by the maximum reductionfactor kt ,max, shown in
Table 7, as follows:
Pm;c ¼ 0:374 � α � d2ffiffiffiffiffiffiffiffiffiffiffiffiffif c
� Ec
qð8Þ
Pm;s ¼ 1:00 � f u � π � d2=4 ð9Þ
kt ¼ 0:7ffiffiffiffiffinrp �bmhp
� hschp
−1� �
≤kt;max ð10Þ
PRm ¼ kt � min Pm;cPm;s�
ð11Þ
where
fc concrete cylinder strengthfu stud tensile strengthEc Young's
modulus of concreted diameter of studbm rib width at mid-height of
the deck profilehp height of the deck ribhsc as-welded height of
studnr number of studs per deck ribemin=2d minimum embedment depth
in the slabα=1.0 for hsc/dN4α ¼ 0:2 � ðhscd þ 1Þ for 3≤hsc/d≤4
However, a basic change of the failuremode, as has been observed
inthe tests, was not considered in EN 1994-1-1 [15,2].
The comparison between the average resistance according to
EN1994-1-1 [2] and Roik et al. [15] and the test results is shown
in Fig.38. It is shown that EN 1994-1-1 generally over-estimates
test results.
For 58 mm deep decking, the width of the rib (bm=81.5 mm) ismuch
smaller than for deckswith comparable heights that are availablein
recent years [1,15]. For tests with 80 mm deep decking, the
embed-ment depth of the head of the stud did not satisfy the
minimum valueof 2 diameters [2]. Because of this, the tested
parameters match thelimits of the database, which has been used in
the calibration of EN1994-1-1.
Table 8Effective area of weld collar according to Konrad
[11].
d hWulst dWulst AWults,eff
[mm] [mm] [mm] [mm2]
10 2.5 13.0 16.313 3.0 17.0 25.516 4.5 21.0 47.319 6.0 23.0
63.022 6.0 29.0 87.025 7.0 40.0 140.0
8.2. Comparison with analytical resistances according to
Lungershausen
According to Lungershausen [1], the shear resistance of the
shearconnector is strongly dependent on the deformation behaviour
of theshear stud itself. The resistance is derived from a
load-bearing mecha-nism of the shear stud with two plastic hinges
according to the plasticdesign theory (see Fig. 2). Accordingly,
themeanvalue for the resistanceof a stud is presented in Eq. (12),
as follows:
PRm ¼ 1:006 � βffiffiffiffiffinrp2Mplã � d ð12Þ
with:
Mpl=σF ⋅d3/6 plastic bending resistance of studσF=500 N/mm 2
nominal steel strengthϞ ¼ 0:8 � ðhpboÞ
2 þ 0:6 relative distance of hingesbo width of rib at its topnr
number of studs per ribemin ¼ 2d
ffiffiffiffiffinr
pminimum embedment depth
β 1:00 for open deck shapes1:10 for re‐entrant deck shapes
�
The results of this comparison are shown in Fig. 39. The
predicted re-sistances are also non-conservative in most cases.
For the test series NR1 with 80 mm deep decking and single
studsper rib, the required embedment depth, emin, was rarely
satisfied butthe shear resistancewas well predicted with a test
load of about 300 kN.
For series 1-04 to 1-06with 58mmdeepdecking, the load-bearing
ca-pacity is overestimated, even though the mechanism with two
plastichinges developed in the studs in the tests. The width of the
decking ismuch smaller than for comparable decks used by
Lungerhausen [1] to de-termine the distance between the plastic
hinges in the stud. This means,that Ϟ in Eq. (12) has not been
calibrated for this narrow type of rib.
The tests with pairs of studs in 80 mm deep decking do not
satisfythe required embedment depth, emin, and a failure mechanism
withonly one plastic hinge developed (see Fig. 20). However, some
studsin tests with higher transverse loading showed plastic
deformations inthe upper stud shank, but the cross-section cannot
be assumed to befully plastic. Thus, the load-bearing capacity is
also overestimated.
8.3. Comparison with analytical resistances according to
Konrad
The third study considers the reduction factors presented by
Konrad[11]. The shear resistance of a stud in solid slabs is
calculated accordingto Eqs. (13) and (14) and is reduced with
reduction factors, which con-sider the geometry of the shear
connection and the welding position.
Accordingly, the studs in 58 mm deep decking are in
theunfavourable position and Eq. (15) is used. For 80 mm deep
decking,the ratio hsc/hp is b1.56 and Eq. (16) is used.
These equations have been derived for through deck-welded
studs.They over-estimate the resistances for series 1-04 to 1-06
and 3-02.This effect may originate out of the use of a pre-punched
decking. TheKonrad formulae are as follows:
Pm;c ¼ 39:5312 � AWulst;eff � f 2=3c þ 3:72 � d2 � f 1=3c � f
1=2u ð13Þ
Pm;s ¼ 38:2959 � AWulst;eff � f 2=3c þ 0:57 � f u�2 ð14Þ
kunfav;3 ¼ kn � 0:317bmhp
þ 0:06� �
≤0:8 ð15Þ
kmid;1 ¼ kn � 6:79 � 10−4bmhp
� �2þ 0:170 bm
hpþ 0:250hsc
hp
" #≤1:0 ð16Þ
where:
-
Table 9Comparison of ratios Pe/PRm of different analysis methods
for the prediction of the shearstud resistances.
EN 1994-1-1 Lungershausen Konrad
All tests μ 0.721 0.846 0.734s 0.112 0.121 0.133V 0.155 0.143
0.181
58 mm-n = 1 μ 0.645 0.763 0.854s 0.031 0.035 0.040V 0.048 0.046
0.046
80 mm-n = 2 μ 0.754 0.871 0.615s 0.130 0.131 0.096V 0.172 0.150
0.155
80 mm-n = 1 μ 0.828 0.996 0.770s 0.048 0.058 0.046V 0.058 0.058
0.060
μ: average of the ratios Pe/PRms: standard deviation of the
ratios Pe/PRmV: coefficient of variation of the ratios Pe/PRm.
352 S. Nellinger et al. / Journal of Constructional Steel
Research 128 (2017) 335–353
AWulst,eff area of weld collar (see Table 8)bm rib width at
mid-height
kn 1:00 for single studs�
0:80 for pairs of studs
The results of the comparison are shown in Fig. 40 and show that
theequations of Konrad [11] over-estimate the resistance of the
push-outtests in this paper.
The predictions according to Konrad [11] for specimens with
58mmdeep decking in series 1-04 to 1-06 are much better than EN
1994-1-1.This indicates the importance of considering the stud
position becausethe equation for studs in an unfavourable
positionwas used to calculatethe shear resistance. The
over-estimation could be because the deckingwas pre-punched prior
to welding the studs.
The predictions for all testswith 80mmdeep decking show the
larg-est deviation. Konrad [11] did not report a comparable failure
mecha-nism for this deck height. Because of this, the reduction
factor kmid,1 inEq. (16) may not be accurate for rib pry-out
failure, even though all re-quirements for the application of this
factor are satisfied.
8.4. Comparison of the presented methods
A comparison of the results for all three analytical methods to
pre-dict the shear stud resistance is given in Table 9 and Fig. 41.
If all 20test results are considered, the method proposed by
Lungershausen[1] is the most accurate. Significant differences
between the methodsof EN 1994-1-1 [2,15] and Konrad [11] were found
when the stud posi-tion is considered. The method by Konrad gives
the most accurate re-sults for tests with 58 mm deep decking, where
the studs are inunfavourable position because of the narrow deck
rib, but predictionsfor tests with 80 mm deep decking are the most
inaccurate of thethree methods.
The results of the comparison show that the procedure of
multiply-ing the resistance of a shear stud in a solid slab by an
empirical derivedreduction factor is insufficient to determine the
resistance of a shearstud in composite slabs with deep profiles.
The empirical reduction fac-tors are only accurate within the range
of parameters, which have beencovered by the database of test
results considered for the calibration ofthe reduction factor, kt.
The tests presented in this paper are not wellcovered by these
databases—in terms of stud position, welding proce-dure and the
observation of rib pry-out failure. The statistical evaluationof
the influencing parameters [11] is not able to reflect
significantchanges in the load-bearing behaviour due to different
failure modes.
Design procedures which are based on the stud shear resistance
de-rived from mechanical models are better able to reflect failure
modes,which were not covered by the statistical evaluation. This
can be seenas the equation proposed by Lungershausen [1] shows the
best results
in the presented comparisons. However, this equation also
over-esti-mates the shear resistances. It is based on a simplified
mechanicalmodel and does not consider the concrete strength, the
welding proce-dure or the stud position.
9. Conclusions
A series of 20 push-out tests with deep steel decking placed
trans-versely to the steel beams was conducted. The investigated
parameterswere as follows
• shape and height of the composite deck profile• number of
reinforcement layers• welding procedure• degree of transverse
loading• influence of eccentric transverse loading.
The results of the presented push-out tests show that the
load–slipbehaviour and the failure mechanisms depend on the
geometry of thedeck profile and the embedment depth of the head of
the studs in theconcrete topping. The significance of the influence
of transverse loadingdepended on the observed failuremechanisms. An
eccentric applicationof the transverse load (to reflect negative
moments in the slab) did notshow amajor influence on the load–slip
behaviour, whichwould be rel-evant for design.
For shear studs with sufficient embedment depth, a load–slip
curvewith two load peaks develops, as described by Lungershausen
[1]. Inthis case, the transverse load influences the load–slip
behaviour similarto an increase in concrete strength, but the
effect is relatively small.
With a small embedment depth, rib pry-out failure occurs at
lowslips. The observed influence of the transverse loading in the
tests wasinconsistent for single studs and pairs of studs per rib.
For pairs ofstuds, the contact pressure at the failure surface was
increased and ledto a higher shear force. For single studs, the
transverse load led to a fail-ure of the compression struts in the
failure cone or failure of the com-pressed face at lower shear
forces.
Based on these results, it is recommended to apply a
concentrictransverse load of 5% of the total test load to the
push-out specimen, ifthe height of the stud is less or equal to
1.56 times the deck height. Oth-erwise, transverse loading is not
required.
The comparisons of the experimental and analytical resistances
arenon-conservative in most cases. The best correlation between the
ex-perimental and analytical resistances was obtained using the
model ac-cording to Lungershausen [1]. The mechanical model is able
to predictthe shear resistance more accurately, even though the
tested parame-ters were outside the considered range used by
Lungershausen [1] inthe calibration of the model.
The presented test results will assist in the development of
im-proved design equations for the load-bearing capacity of stud
shear con-nectors, where the additional failure mechanism rib
pry-out, theinfluence of the transverse loading as well as the
position of the shearstuds have to be considered.
Acknowledgements
The research leading to these results is part of a common
project ofthe Steel Construction Institute, University of
Stuttgart, University ofLuxembourg, University of Bradford and
ArcelorMittal and has receivedfunding from European Communitys
Research Fund for Coal and Steel(RFCS) under grant agreement no
[RFCS-CT-2012-00030].
References
[1] H. Lungershausen, Zur Schubtragfähigkeit von
Kopfbolzendübeln(Ph.D. thesis)Institut für konstruktiven
Ingenieurbau, Ruhr-Universität Bochum, 1988.
[2] DIN EN 1994-1-1: 2010–12: Eurocode 4: Bemessung und
Konstruktion vonVerbundtragwerken aus Stahl und Beton - Teil 1–1:
Allgemeine Bemessungsregeln
http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0005http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0005
-
353S. Nellinger et al. / Journal of Constructional Steel
Research 128 (2017) 335–353
und Anwendungsregeln für den Hochbau; Deutsche Fassung EN
1994-1-1:2004 + AC:2009, 2010-12.
[3] K. Roik, G. Hanswille, Zur Dauerfestigkeit von
Kopfbolzendübeln beiVerbundträgern, Bauingenieur 62 (1987)
273–285.
[4] S. Hicks, A. Smith, Stud shear connectors in composite beams
that support slabs withprofiled steel sheeting, Struct. Eng. Int.
2/2014 (2014) 246–253.
[5] P. Döinghaus, Zum Zusammenwirken Hochfester Baustoffe in
Verbundträgern(Ph.D.thesis) RWTH Aachen, 2001.
[6] S. Ernst, Factors Affecting the Behaviour of the Shear
Connection of Steel-ConcreteComposite Beams(Ph.D. thesis)
University of Western Sydney, November 2006.
[7] S. Nellinger, C. Odenbreit, M. Lawson, Push-out tests with
modern deck sheeting andrealistic transverse loading, 12th
International Conference on Steel, Space and Com-posite Structures:
Conference Proceedings, 2014.
[8] D. Lam, Composite Steel Beams Using Precast Concrete Hollow
Core FloorSlabs(Ph.D. thesis) University of Nottingham, March
1998.
[9] D. Lam, J. Qureshi, J. Ye, Effect of shear connector spacing
and layout on the shearconnector capacity in composite beams, J.
Constr. Steel Res. 67 (2011) 706–719.
[10] M.D. Rambo-Roddenberry, Behaviour and Strength of Welded
Stud ShearConnectors(Ph.D. thesis) Virginia Polytechnic Institute
and State University, 2002.
[11] M. Konrad, Tragverhalten von Kopfbolzen in Verbundträgern
bei senkrechtspannenden Trapezblechen(Ph.D. thesis) Institut für
Konstruktion und Entwurf,Stahl- Holz- und Verbundbau, Universität
Stuttgart, 2011.
[12] S. Nellinger, On the Behaviour of Shear Stud Connections in
Composite Beams withDeep Decking(Ph.D. thesis) University of
Luxembourg, 2015.
[13] F. Eggert, S. Nellinger, U. Kuhlmann, C. Odenbreit,
Push-out tests with modern decksheeting to evaluate shear connector
resistances, EUROSTEEL 2014: Conference Pro-ceedings, 2014.
[14] H.B. Kupfer, Das Verhalten des Betons unter mehrachsiger
Kurzzeitbelastung unterbesonderer Berücksichtigung der zweiachsigen
Beanspruchung, Tech. rep.,Deutscher Ausschuss für Stahlbeton, Heft
229, Beuth Verlag, 1973.
[15] K. Roik, G. Hanswille, O. Cunze, O. Lanna, Report on
Eurocode 4—Clause 6.3.2: StudConnectors, Report ec4/8/88, Tech.
Rep. Ruhr-Universität Bochum, 1989.
http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0010http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0010http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0015http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0015http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0020http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0020http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0025http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0025http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0030http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0030http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0030http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0035http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0035http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0040http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0040http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0045http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0045http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0050http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0050http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0050http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0055http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0055http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0060http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0060http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0060http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0065http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0065http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0065http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0070http://refhub.elsevier.com/S0143-974X(16)30265-6/rf0070
Influence of transverse loading onto push-out tests with deep
steel decking1. Introduction1.1. Load-bearing behaviour of shear
connectors1.2. Test setups to investigate the load–slip
behaviour
2. Consideration of transverse loading in the presented push-out
tests3. Test programme and setup3.1. Test setup for transverse
loading3.2. Test programme and material properties
4. Observed load–slip curves and failure modes4.1. General
results of tests with 58mm deep decking4.2. General results of
tests with 80mm deep decking
5. Evaluation of the test results according to EN 1994-1-1 Annex
B26. Discussion of influencing parameters6.1. Influence of variable
versus constant transverse loading6.2. Considerations on the
multi-axial stress state for 58mm deep decking6.3. Influence of the
degree of concentric transverse loading for 80mm deep decking6.4.
Influence of the eccentricity of transverse loads with 58mm deep
decking6.5. Influence of the eccentricity of transverse loads with
80mm deep decking6.6. Influence of through deck welded studs with
80mm deep decking6.7. Influence of the number of reinforcement
layers6.8. Influence of the number of shear studs per rib
7. Proposed testing procedure for push-out specimens with
composite slabs8. Comparison with analytical resistances8.1.
Comparison with analytical resistances according to EN 1994-1-1 and
Roik et al.8.2. Comparison with analytical resistances according to
Lungershausen8.3. Comparison with analytical resistances according
to Konrad8.4. Comparison of the presented methods
9. ConclusionsAcknowledgementsReferences